US 7237075 B2
A persistent snapshot is taken and maintained in accordance with a novel method and system for extended periods of time using only a portion of a computer readable medium of which the snapshot is taken. Multiple snapshots can be taken in succession at periodic intervals and maintained practically indefinitely. The snapshots are maintained even after powering down and rebooting of the computer system. The state of the object of the snapshot for each snapshot preferably is accessible via a folder on volume of the snapshot. A restore of a file or folder may be accomplished by merely copying that file or folder from the snapshot folder to a current directory of the volume. Alternatively, the entire computer system may be restored to a previous snapshot state thereof. Snapshots that occurred after the state to which the computer is restored are not lost in the restore operation. Different rule sets and scenarios can be applied to each snapshot. Furthermore, each snapshot can be written to within the context of the snapshot and later restored to its pristine condition. Software for implementing the systems and methods of snapshots in accordance with the present invention may comprise firmware of a hard disk drive controller or a disk controller board or within the HDD casing itself. The present invention further comprises novel systems and methods in which the systems and methods of taking and maintaining snapshots are utilized in creating and managing temporal data stores, including temporal database management systems. The implications for data mining and exploration, data analysis, intelligence gathering, and artificial intelligence (just to name a few areas) are profound.
1. A method of preserving data of a computing system, comprising:
maintaining multiple non-overlapping snapshots of a volume of data in data storage over a period of time;
determining whether a granule of the volume requires caching prior to being overwritten;
saving the granule of the volume prior to being overwritten if it needs caching to a cache associated with a current snapshot; and
terminating the cache associated with the current snapshot just prior to the start of a next snapshot, thereby caching preserved data with a minimum of redundancy.
2. A computer-readable medium having computer-readable instructions for performing the method of
3. A computer configuration comprising computer-readable medium having computer-readable instructions for performing the method of
This is a continuation-in-part patent application that claims the benefit under 35 U.S.C. §120 to the filing dates of: U.S. nonprovisional patent application Ser. No. 10/248,483, titled, “Persistent Snapshot Management System,” filed Jan. 22, 2003, now abandoned which is a nonprovisional of U.S. provisional patent application No. 60/350,434, titled, “Persistent Snapshot Management System,” filed Jan. 22, 2002; and U.S. nonprovisional patent application Ser. No. 10/349,474, titled, “Persistent Snapshot Management System,” filed Jan. 22, 2003, now abandoned which is a nonprovisional of U.S. provisional patent application No. 60/350,434, titled, “Persistent Snapshot Management System,” filed Jan. 22, 2002. Each of these U.S. patent applications is hereby incorporated herein by reference.
Program Source Code Code.txt includes 83598 lines of code representing an implementation of a preferred embodiment of the present invention. The programming language is C++ and is intended to run on the Windows 2000 operating system. This program source code is incorporated herein by reference as part of the disclosure.
Data of a computer system generally is archived on a periodic basis, such as at the end of each day; at the end of each week; at the end of each month; and/or at the end of each year. Data may also be archived before or after certain events or actions. When archived, the data is logically consistent, i.e., all of the data subjected to the archiving process at any point in time is maintained in the state as it existed at that particular point in time.
The archived data provides a means for restoring a computer system to a previous, known state, which may be necessary when performing disaster recovery such as occurs when data in a primary storage system is lost or corrupted. Data may be lost or corrupted if the primary storage system, such as a hard disk drive or other mass storage system, is physically damaged, if the operating system of the primary storage system crashes, or if files of the primary storage system are infected by a computer virus. By archiving the data on a periodic basis, the computer system always can be restored to its state as it existed at the most recent backup time, thereby minimizing any permanent data loss should disaster recovery actually be performed. The restoration may be of one or more files of the computer system or of the entire computer system itself.
There are numerous types of methods for archiving data. One type includes the copying of the data subject to the archive to a backup storage system. Typically, the backup storage system includes backup medium comprising magnetic computer tapes or optical disks used to store backup copies of large amounts of data, as is often associated with computer systems. Furthermore, each backup tape or optical disk can be maintained in storage indefinitely by sending it offsite. In order to minimize costs, such tapes and disks also can be reused on a rolling basis if such backup medium is rewriteable, or destroyed if not rewriteable and physical storage space for the backups is limited. In this later scenario, the “first in-first out” methodology is utilized in which the tape or disk having the oldest recording date is destroyed first.
One disadvantage to archiving data by making backups is that the data subject to the archiving process is copied in totality onto the backup medium. Thus, if 250 gigabytes of data is to be archived, then 250 gigabytes of storage capacity is required. If a terabyte of data is to be backed up, then a terabyte of storage capacity is required. Another related disadvantage is that as the amount of data to be archived increases, the period of time required to perform the backup increases as well. Indeed, it may take weeks to archive onto tape a terabyte of data. Likewise, it may take weeks if it becomes necessary to restore such amount of data.
Yet another disadvantage is that sometimes an “incremental” backup is made, wherein only the new data that has been written since the last backup is actually copied to the backup medium. This is in contrast to the “complete” backup of the data, wherein all the data subject to the archiving process is copied whether or not it is new. Restoring archived data from complete and incremental backups requires copying from a complete backup and then copying from the incremental backups thereafter made between the time point of the complete backup until the time point of the restoration. A fourth and obvious disadvantage is that when the backup medium in the archiving process is stored offline, the archived data must be physically retrieved and mounted for access and, thus, is not readily available on demand.
In view of the foregoing, it will be apparent that it is extremely inefficient to utilize backups for restoring data when, for example, only a particular user file or some other limited subset of the backup is required. To address this concern, a snapshot can be taken of data whereby an image of the data at the particular snapshot moment can later be accessed. The object of the snapshot for which the image is provided may be of a file, a group of files, a volume or logical partition, or an entire storage system. The snapshot may also be of a computer-readable medium, or portion thereof, and the snapshot may be implemented at the file level or at the storage system block level. In either case, the data of the snapshot is maintained for later access by (1) saving snapshot data before replacement thereof by new data in a “copy on write” operation, and (2) keeping track of all the snapshot data, including the snapshot data still residing in the original location at the snapshot moment as well as the snapshot data that has been saved elsewhere in the copy-on-write operation. Typically, the snapshot data that is saved in the copy-on-write operation is stored in a specially allocated area on the same storage medium as the object of the snapshot. This area typically is a finite storage data of fixed capacity.
The use of snapshots has advantages over the archiving process because a backup medium separate and apart from a primary storage medium is not required, and the snapshot data is stored online and, thus, readily accessible. A snapshot also only requires storage capacity equal to that amount of data that is subjected to the copy-on-write operation; thus, all of the snapshot data need not be saved to a specifically allocated data storage area if all of the snapshot data is not to be replaced. The taking of a snapshot also is near instantaneous.
Advantageously, a snapshot may also be utilized in creating a backup copy of a primary storage medium onto a backup medium, such as a tape. As disclosed, for example, in Ohran U.S. Pat. No. 5,649,152, a snapshot can be taken of a base “volume” (a/k/a a “logical drive”), and then a tape backup can be made by reading from and copying the snapshot onto tape. During this archive process, reads and writes to the base volume can continue without waiting for completion of the archive process because the snapshot itself is a non-changing image of the data of the base volume as it existed at the snapshot moment. The snapshot in this instance thus provides a means by which data can continue to be read from and written to the primary storage medium while the backup process concurrently runs. Once the backup is created, the snapshot is released and the resources that were used for taking and maintaining the snapshot are made available for other uses by the computer system.
A disadvantage to utilizing snapshots is that a snapshot is not a physical duplication of the data of the object of the snapshot onto a backup medium. A snapshot is not a backup. Furthermore, if the storage medium on which the original object of the snapshot resides is physically damaged, then both the object and the snapshot can be lost. A snapshot, therefore, does not provide protection against physical damage of the storage medium itself.
A snapshot also requires significant storage capacity if it is to be maintained over an extended period of time, since snapshot data is saved before being replaced and, over the course of an extended period of time, much of the snapshot data may need saving. The storage capacity required to maintain the snapshot also dramatically increases as multiple snapshots are taken and maintained. Each snapshot may require the saving of overlapping snapshot data, which accelerates consumption of the storage capacity allocated for snapshot data. In an extreme case, each snapshot ultimately will require a storage capacity equal to the amount of data of its respective object. This is problematic as the storage capacity of any particular storage medium is finite and, generally, the finite data storage will not have sufficient capacity to accommodate this, leading to failure of the snapshot system.
Accordingly, snapshots generally are used solely for transient applications, wherein, after the intended purpose for which the snapshot is taken has been achieved, the snapshot is released and system resources freed, perhaps for the provision of a subsequent snapshot. Furthermore, because snapshots are only needed for temporary purposes, the means for tracking the snapshot data may be stored in RAM memory of a computer and is lost upon the powering down or loss of power of the computer, and, consequently, the snapshot is lost. In contrast thereto, backups are used for permanent data archiving.
Accordingly, a need exists for an improved system and method that, but for protection against physical damage to the storage medium itself, provides the combined benefits of both snapshots and backups without the time and storage capacity constraints associated with snapshots and backups. One or more embodiments of the present invention meet this and other needs, as will become apparent from the detailed description thereof below and consideration of the computer source code incorporated herein by reference and disclosed in the incorporated provisional U.S. patent application.
Briefly described, the invention comprises a snapshot management system.
Further features and benefits of the present invention will be apparent from a detailed description of preferred embodiments thereof taken in conjunction with the following drawings, wherein similar elements are referred to with similar reference numbers, and wherein,
As a preliminary matter, it will readily be understood by those persons skilled in the art that the present invention is susceptible of broad utility and application in view of the following detailed description of preferred embodiments of the present invention. Many devices, methods, embodiments, and adaptations of the present invention other than those herein described, as well as many variations, modifications, and equivalent arrangements thereof, will be apparent from or reasonably suggested by the present invention and the following detailed description thereof, without departing from the substance or scope of the present invention. Accordingly, while the present invention is described herein in detail in relation to preferred embodiments, it is to be understood that this disclosure is illustrative and exemplary and is made merely for purposes of providing a full and enabling disclosure of preferred embodiments of the invention. The disclosure herein is not intended nor is to be construed to limit the present invention or otherwise to exclude any such other embodiments, adaptations, variations, modifications and equivalent arrangements, the present invention being limited only by the claims appended hereto or presented in any continuing application, and the equivalents thereof.
Exemplary Operating Environment
With reference to
A number of program modules may be stored in the drives and RAM 25, including an operating system 35, one or more application programs 36, the Persistent Storage Manager (PSM) module 37, and program data 38. A user may enter commands and information into the computer 20 through a keyboard 40 and pointing device, such as a mouse 42. Other input devices (not shown) may include a microphone, joystick, game pad, satellite dish, scanner, or the like. These and other input devices are often connected to the processing unit 21 through a serial port interface 46 that is coupled to the system bus 23, but may be connected by other interfaces, such as a game port or a universal serial bus (USB). A monitor 47 or other type of display device is also connected to the system bus 23 via an interface, such as a video adapter 48. In addition to the monitor 47, computers typically include other peripheral output devices (not shown), such as speakers or printers.
The computer 20 may operate in a networked environment using logical connections to one or more remote computers, such as a remote computer 49. The remote computer 49 may be a server, a router, a peer device, or other common network node, and typically includes many or all of the elements described relative to the computer 20, although only a memory storage device 50 has been illustrated in
When used in a LAN networking environment, the computer 20 is connected to the LAN 51 through a network interface 53. When used in a WAN networking environment, the computer 20 typically includes a modem 54 or other means for establishing communications over the WAN 52, such as the Internet. The modem 54, which may be internal or external, is connected to the system bus 23 via the serial port interface 46. In a networked environment, program modules depicted relative to the computer 20, or portions thereof, may be stored in the remote memory storage device. It will be appreciated that the network connections shown are exemplary and other means of establishing a communication's link between the computers may be used.
Exemplary Snapshot System
Turning now to
Such system includes components of a computer system, such as an operating system 210. The system also includes a persistent storage manager (PSM) module 220, which performs methods and processes of the present invention, as will be explained hereinafter. The system also includes at least one finite data storage medium 230, such as a hard drive or hard disk. The storage medium 230 comprises two dedicated portions, namely, a primary volume 242 and a cache 244. The primary volume 242 contains active user and system data 235. The cache 244 contains a plurality of snapshot caches 252, 254, 256 generated by the PSM module 220.
The operating system 210 includes system drivers 212 and a plurality of mounts 214, 216, 218. The system also includes a user interface 270, such as a monitor or display. The user interface 270 displays snapshot data 272 in a manner that is meaningful to the user, such as by means of conventional folders 274, 276, 278. Each folder 274, 276, 278 is generated from a respective mount 214, 216, 218 by the operating system 210. Each respective folder preferably displays snapshot information in a folder and file tree format 280, as generated by the PSM module 220. Specifically, as will be discussed in greater detail herein, the PSM module 220 in conjunction with the operating system 210 is able to display current and historical snapshot information by accessing both active user and system data 235 and snapshot caches 252, 254, 256 maintained on the finite data storage medium 230.
Methods and further processes for taking, maintaining, managing, manipulating, and displaying snapshot data according to the present invention will be described in greater detail hereinafter.
Exemplary Disk Level Operations
Referring generally to
The second section 320 of
The third section 330 of
The fourth section 340 of
First, like sections 320 and 330, each snapshot specific cache 342,344,346 is illustrated as a grid, with rows 1 through 4 corresponding to volume address locations 1 through 4 and with each column corresponding to a respective point in time from the timeline in section 310. Each grid shows how each respective snapshot specific cache is populated over time. Specifically, it should be understood that a snapshot specific cache comprises potential granules corresponding to each row of address locations of the volume but only for point of time beginning when the respective snapshot is taken and ending with the last point of time just prior to the next succeeding snapshot. There is no overlap in points of time between any two snapshot specific caches.
Thus, each snapshot specific cache grid 342,344,346 identifies what data has been recorded to that respective cache and when such data was actually recorded. For example, as shown in the first snapshot specific cache grid 342, data “C” is written to address 3 at time 8 and is maintained in that address for this first cache thereinafter. Likewise, data “D” is written to address 4 at time 9 and maintained at that address for this first cache thereinafter. Correspondingly, in the second snapshot specific cache 344, data “G” is written to address 4 at time 14 and maintained at that address for this second cache thereinafter. In the third snapshot specific cache 346, data “A” is written to address 1 at time 21 and maintained at that address for this third cache thereinafter, data “F” is written to address 3 at time 20 and maintained at that address for this third cache thereinafter, and data “I” is written to address 4 at time 20 and maintained at that address for this third cache thereinafter. The shaded granules in each of the snapshot specific cache grids 342,344,346 merely indicate that no data was written to that particular address at that particular point in time in that particular snapshot specific cache; thus, no additional memory of the data storage medium is used or necessary.
The second manner of illustrating each snapshot specific cache is shown by column 350, which includes the first snapshot specific cache 352, the second snapshot specific cache 354, and the third snapshot specific cache 356. As explained previously, each snapshot specific cache only comprises potential granules corresponding to each row of address locations of the volume for points of time beginning when the respective snapshot is taken and ending with the last point of time just prior to the next succeeding snapshot. In other words, the first snapshot cache was being dynamically created between times 5 and 10 and actually changed from time 8 to time 9; however, at time 11, when the second snapshot was taken, the first snapshot cache became permanently fixed, as shown by cache 352. Likewise, the second snapshot cache was being dynamically created between times 11 and 17 and actually changed from time 13 to time 14; however, at time 18, when the third snapshot was taken, the second snapshot cache became permanently fixed, as shown by cache 354. Finally, the third snapshot cache is still in the process of being dynamically created beginning at time 18, and changed from time 19 to time 20 and from time 20 to time 21; however, this cache 356 will not actually become fixed until a fourth snapshot (not shown) is taken at some point in the future. Thus, even though cache 356 has not yet become fixed, it can still be accessed and, as of time 22, contains the data as shown.
Further, it should be understood that the shaded granules in each of the snapshot specific caches 352,354,356 merely indicate that no data was written or has yet been written to that particular address when that particular cache was permanently fixed in time (for caches 352, 354) or as of time 22 (for cache 356); thus, no additional memory of the data storage medium has been used or was necessary to create the caches 352,354,356. Stated another way, only the data shown in the fifth section of
Although it should be self evident from
Now, proceeding with the time point by time point analysis of
However, at time 3, a command to write data “E” to address 2 is received.
Data “E” is written to this address at time 4, replacing data “B.” Data “B” is not written to any snapshot cache because no snapshots have yet been taken of the volume. Thus, at time 5, when the first snapshot is taken, the values of the volume are “AECD.” It should be noted that although the snapshot has been taken at time 5, there is no need, yet, to record any of the data in the volume to snapshot cache because the current volume accurately reflects what the state of the volume is or was at time 5. Since the volume is still the same as it was at time 5, nothing changes at time 6.
At time 7, a command to write data “F” to address 3 is received. Data “F” will be replacing data “C” on the volume; however, because data “C” is part of snapshot 1, data “F” is not immediately written to this address. First, data “C” must be written to the first snapshot cache, as shown at time 8 in cache grid 342. Once data “C” has been written to the first snapshot cache, data “F” can then be safely written to address 3 of the volume, which is shown at the next time point, time 9. This process is generally described as the “copy on write” process in conventional snapshot parlance. The copy on write process is repeated for writing data “G” to the volume and writing data “D” to the first snapshot cache but it is staggered in time from the previous copy on write process.
The second snapshot is taken at time 11. The volume at that point is “AEFG.” Again, as stated previously, it is at this point that the first snapshot cache 342 is permanently fixed, as shown by granules 352. It is no longer necessary to add any further information to this first snapshot cache 352.
The third snapshot is taken at time 18. The volume at that point is now “AEFI.” Again, as stated previously, it is at this point that the second snapshot cache 344 is permanently fixed, as shown by granules 354. It is no longer necessary to add any further information to this second snapshot cache 354.
At time 19, commands to write data “J” to address 3 and data “K” to address 4 are received. Data “J” will be replacing data “F” and data “K” will be replacing data “I”; however, because data “F” is part of snapshots 1 and 2 and because data “I” was part of snapshot 2, data “J” and “K” are not immediately written to these addresses. The copy on write process is performed for each address so that data “F” and “I” are written to the third snapshot cache at time 20 as shown in grid 346. Once this has occurred, data “J” and “K” can be safely written to addresses 3 and 4, respectively, of the volume at time 21. These particular copy on write procedures are included so that one can easily see the different state of the cache for addresses 3 and 4 for each different snapshot cache 352,354,356. Specifically, it was not necessary to include data “F” as part of the second snapshot cache 354, even though it was on the volume at the time of the second snapshot.
Finally, at time 20 a command to write data “L” to address 1 is received. Data “L” will be replacing data “A”; however, because data “A” is part of snapshots 1, 2, and 3, data “L” is not immediately written to this address. The copy on write process is performed so that data “A” is written to the third snapshot cache at time 21 as shown in grid 346. Once data “A” has been written to the third snapshot cache, data “L” can be safely written to address 1 of the volume at time 22. This particular copy on write procedure is included herein to illustrate that, even though data “A” was part of snapshots 1 and 2, it did not need to be written to cache until it was actually replaced. Further, it is not necessary to copy data “A” to the first or second snapshot caches 352,354 it only needs to be part of the third snapshot cache 356. Again, the third snapshot cache 356 will becomes fixed as soon as the next snapshot is taken.
Finally, it should be noted that data “E,” which is part of all three snapshots is not written to cache because it is never replaced during the time duration of
Turning briefly now to
Unlike the present invention, each snapshot cache 442, 444, and 446 begins at its respective time of snapshot (time 5, 11, and 18, respectively) but then continues ad infinitum, as long as the system is maintaining snapshots in memory, rather than stopping at the point in time just prior to the next snapshot being taken. The result of this is that the same data is recorded redundantly in each snapshot cache 452, 454, and 456. For example, data “A” is stored not only in the third snapshot cache 456 at address 1 but also at address 1 in the first and second snapshot caches 452,454, respectively. Likewise, data “F” is stored not only in the third snapshot cache 456 at address 3 but also in the second snapshot cache 454 also at address 3. The redundancy of this prior art system is illustrated as well with reference to table 460, which may be contrasted easily with table 360 in
Turning now to
If the determination in Step 520 is negative, then the system determines (Step 550) whether a command to write new data to the volume has been received. If not, then the system returns to Step 510 to wait for another command. If so, then the system determines (Step 560) whether the data on the volume that is going to be overwritten needs to be cached. For example, from
Turning now to
Turning first to
Column 637 identifies what data was contained in the volume at time 5, when the first snapshot was taken; however, it is assumed that the system only has access to the data from the current volume 635, as it exists immediately after time 22, and to the snapshot caches 652, 654, and 656. Column 670 represents what the system would read as the image of the first snapshot. Thus, after the proper procedures are performed, column 670 should match column 637.
To determine the data on the volume at the first snapshot, it is first necessary to examine the first snapshot cache 652. Each separate address granule is examined and, if any granule has any data therein, such data is represented in column 670 and would be read by the system as part of the first snapshot. As shown, the first snapshot cache has data “C” at address 3 and data “D” at address 4. These are represented in column 670 at addresses 3 and 4 respectively.
Next, each address granule for which data has not yet been determined are considered. Thus, addresses 1 and 2 are considered, but addresses 3 and 4 are not considered because values have been determined for those addresses. Accordingly, the second snapshot cache 654 is then examined in an attempt to determine values for addresses 1 and 2. If either address has data found in the second snapshot cache 654, then such data is represented in column 670 at its respective address. As illustrated in
This process is repeated for each successive snapshot cache until all successive snapshot caches have been considered or until no value for any address remains undetermined. As shown, addresses 1 and 2 of the third snapshot cache 656 are next examined, and data “A” from address 1 in the third snapshot cache 656 is found and thus represented in address 1 of column 670.
Once all snapshot caches have been examined, any addresses for which no data was found from such snapshot caches is obtained directly from the relevant address(es) of the current volume 635. In this case, data “E” from the current volume at address 2 is represented in column 670 as the value for address 2.
As shown, the data 637 as it existed in the volume at time 5 is correctly represented in column 670 by following the above process.
Likewise, turning to
Turning now to
The second section 720 of
The third section 730 of
The fourth section 740 of
Further, it should be understood that the shaded granules in each of the snapshot specific caches 752, 754, 756 merely indicate that no data was written or has yet been written to that particular address when that particular cache was permanently fixed in time (for caches 752, 754) or as of time 20 (for cache 756); thus, no additional memory of the data storage medium has been used or was necessary to create the caches 752, 754, 756. Stated another way, only the data shown in the fifth section of
Although it should be self evident from
Now, proceeding with the time point by time point analysis of
At time 2, commands to write data “E” to address 2 and data “F” to address 3 are received. At time 3, a command to write data “G” to address 4 is received. Data “E” is written to address 2 at time 3, replacing data “b”; data “F” is written to address 3 also at time 3, replacing data “c”; and data “G” is written to address 4 at time 4, replacing data “d.” Data “b” and “c” and “d” are not written to any snapshot cache for either of two reasons: they are lower case, which means they are undesirable and do not need to be cached, and they have been overwritten prior to the first snapshot and thus do not get cached.
At time 4, a command to write data “H” to address 3 is received. Data “H” is written to address 3 at time 5, replacing data “F.” It should be noted that data “H” merely replaces data “F” in the volume.
At time 4, a command to delete the data stored at address 2 is received. Thus, data “E” becomes data “e” at time 4 in the volume. Thus, at time 6, when the first snapshot is taken, the values of the volume are “aeHG.” Data “H” and “G” are now “primed,” as denoted by the prime symbol to indicate that such data should be written to cache if they are ever overwritten by different data. As will become apparent, it is not necessary to write such data to cache if it is merely designated for deletion because it will still be accessible at its respective address location on the volume until it is actually overwritten.
It should be noted that although the snapshot has been taken at time 6, there is no need, yet, to record any of the (upper case) data in the volume to snapshot cache because the current volume accurately reflects what the state of the volume is or was at time 6. Since the volume is still the same as it was at time 6, nothing changes at time 7.
The second snapshot is taken at time 11. The volume at that point is “aeIG.” Data “I” is now “primed,” as denoted by the prime symbol, and data “G” remains primed. Again, as stated previously, it is at this point that the first snapshot cache 752 is permanently fixed. It is no longer necessary to add any further information to this first snapshot cache 742.
At time 13, a command to delete the data stored at address 4 is received. Thus, data “G” becomes data “g” in the volume at time 13. The third snapshot is taken at time 15. The volume at that point is “aeIg.” Data “I” remains “primed” and data “g,” although now designated as ready for deletion, also remains primed. Again, as stated previously, it is at this point that the second snapshot cache 754 is permanently fixed (with no data stored therein). It is no longer necessary to add any further information to this second snapshot cache 744.
Then, at time 17, a command to write data “J” to address 4 is received. Data “J” will be replacing data “g,” again, which has already been designated for deletion. However, because data “g” was part of both snapshots 2 and 3, data “J” is not immediately written to this address. The copy on write process is performed so that data “G” is written to the third snapshot cache at time 18 as shown in grid 746. Once data “G” has been written to the third snapshot cache 746, data “J” can be safely written to address 4 of the volume at time 19.
Finally, it should be noted that data “I,” which is part of two of the snapshots remains primed because it has not yet been overwritten and, thus, has not yet been written to cache during the time duration of
Turning briefly to
Turning now to
If the determination in Step 920 is negative, then the system determines (Step 950) whether a command to write new data to the volume has been received. If so, then the system determines (Step 960) whether the data on the volume that is going to be overwritten needs to be cached (i.e., has the data been “primed” ?). For example, from
If the determination in Step 950 is negative, then the system determines (Step 990) whether a command to delete data from the volume has been received. If not, then the process returns to Step 910 to wait for another command. If so, then the system designates or indicates (Step 995) that the particular volume data can be deleted and the associated space on the volume is available for new data. The process then returns to Step 910 to wait for another command.
Turning now to
The process of creating a modified first historical volume 1070 then is quite similar to the process of recreating an actual historical volume, as illustrated by column 670 from
The process of creating the modified first historical volume, however, starts first with the write snapshot cache corresponding to the snapshot to which the system is being reverting. In
Likewise, turning to
Turning briefly to
Specifically, the system waits (Step 1110) for a request to replace a block of data on the volume. Step 1110 is triggered, for example, when a command to write old data to cache is received (as occurs in Step 570 of
The system then checks (Step 1120) to determine whether a fault has occurred. If so, the system indicates (Step 1170) that there has been a failure, and the write on copy process is halted. If the determination in Step 1120 is negative, then the system writes (Step 1125) the old or primed data to the current snapshot cache.
Again, the system then checks (Step 1130) to determine whether a fault has occurred. If so, the system indicates (Step 1170) that there has been a failure, and the copy on write process is halted. If the determination in Step 1130 is negative, then the system determines (Step 1135) whether the snapshot cache is temporary. If so, then the system merely writes (Step 1150) an entry to the memory index. If the snapshot cache is not temporary, then the system writes (Step 1140) an entry to the disk index file.
Again, the system then checks (Step 1145) to determine whether a fault has occurred. If so, the system indicates (Step 1170) that there has been a failure, and the copy on write process is halted. If the determination in Step 1145 is negative, then the system also writes (Step 1150) an entry to the memory index.
Finally, the system again checks (Step 1155) to determine whether a fault has occurred. If so, the system indicates (Step 1170) that there has been a failure, and the copy on write process is halted. If the determination in Step 1155 is negative, then the system indicates (Step 1160) that the write to the cache was successful and the system then allows the new data to be written to the volume over the old data that was cached.
As will be apparent from the foregoing detailed description, this preferred embodiment of a method of the present invention provides a means for taking and maintaining a snapshot that is highly efficient in its consumption of the finite storage capacity allocated for the snapshot data, even when multiple snapshots are taken and maintained over extended periods of time.
Exemplary System Administrator and User Interfaces
Before continuing with the detailed description of further aspects, systems and methodologies of the present invention, it will be useful to quickly examine a number of system administrator and system user interfaces, in
Turning first to
Hide and Unhide
In accordance with a feature of a preferred method and system of the present invention, a volume address may be omitted from future snapshots, or hidden, as indicated by “−” in
Tracking of Snapshot Data
Snapshot data is tracked in order for the correct granule to be returned in response to reads from the snapshot. A logical structure for tracking snapshot data is illustrated in
Snapshot Delete and Cache Scavenge
In another aspect of the present invention, there may be times when it is necessary or desirable to delete snapshots being maintained by the system of the present invention. Snapshot deletion requires some actions that are not required in less sophisticated systems. Since each snapshot may contain data needed by a previous snapshot, simply releasing the index entries (which are typically used to find data stored on the volume or in cache), and “freeing up” the cache granules associated with the snapshot, may not work. As will be recalled from the above discussions, it is sometime necessary to consult different snapshot caches when trying to read a particular snapshot; thus, there is a need for a way to preserve the integrity of the entire system when deleting undesired snapshots.
The present invention processes such deletions in two phases. First, when a snapshot is to be deleted, the snapshot directory is unlinked from the host operating system, eliminating user access. The Snapshot Master record and each associated Snapshot Entry are then flagged as deleted. Note that this first phase does not remove anything needed by a previously created snapshot to return accurate data.
The second, or “scavenger,” phase occurs immediately after a snapshot is created, a snapshot is deleted, and a system restart. The scavenger phase reads through all Snapshot Entries locating snapshots that have been deleted. For each snapshot entry that has been deleted, a search is made for all data granules associated with that snapshot that are not primed or required by a previous snapshot. Each such unneeded granule is then released from the memory index, the Index File, and the cache file. Other granules that are required to support earlier snapshots remain in place.
When the scavenger determines that a deleted snapshot entry contains no remaining cache associations, it is deleted. When the last snapshot entry associated with a snapshot master entry is deleted, the snapshot master is deleted.
Persistence: Snapshot Reconstruction
In another aspect of the present invention, when the system computer is restarted after a system shutdown (whether intentional or through a system failure), the Header and Index files are used to reconstruct the dynamic snapshot support memory contents.
On restart, the memory structures are set to a startup state. In particular, a flag is set indicating that snapshot reconstruction is underway, the primed map is set with all entries primed, and the cache granule map set to all entries unused. The Header File is then consulted to create a list of Snapshot Master entries, Snapshot Entries, and address of the next available cache file granule.
During the remainder of the reconstruction process, writes may occur to volumes that have active snapshots. Prior to completion of snapshot reconstruction, granule writes to blocks that are flagged prime are copied to the end of the Cache file and recorded in the memory index. The used cache granule map and next available granule address are likewise updated. One skilled in the art will appreciate that setting the prime table to all primed and writing only to the end of the granule cache file will record all first writes to the volume. At this phase, some redundant data is potentially preserved while the prime granule map is being recreated.
Each index entry is consulted in creation order sequence. Blank entries, entries that have no associated Snapshot Entry, and entries that are not associated with a currently available volume device are ignored. Each other entry is recorded in the memory index. If any duplicate entries are located, the subsequently recorded entry replaces the earlier entry. An entry is considered a duplicate if it records the same snapshot number, volume granule address, and cache granule address. The age of each index entry is indicated by a time stamp or similar construct when the entry was originally created.
At this stage in reconstruction, the index in memory is completed. Each snapshot will then be consulted to create the single system wide primed granule map and used cache map.
For each memory index entry for the snapshot the associated primed granule map element is cleared and the granule cache map entry set.
On completion the flag indicating snapshot reconstruction is reset. The cache granule map, primed map, memory index, and file index have been restored to include the state at shutdown, as well as all preserved volume writes that occurred during the reconstruction process.
Restoration of System to Another State
A preferred embodiment of the present invention also provides restore functionality that allows restoration of a volume to any state recorded in a snapshot while retaining all snapshots. This is accomplished by walking through the index while determining which granules are being provided by the cache for the restored snapshot. Those volume granules are replaced by the identified granules from cache. This replacement operation is subject to the same volume protection as any other volume writes, so the volume changes engendered by the restore are preserved in the snapshot set.
The operation begins at Step 3702 when a restore command is received. In Step 3704 a loop through all volume granule addresses on the system is prepared. At Step 3706 the next volume granule address is read. At Step 3708 a process restores the selected granule by searching for the selected granule in each snapshot index commencing with the snapshot to be restored (Step 3712) and ending with the most recent snapshot (Step 3716). The process 12 and 3714 establishes index and end counters to traverse the snapshots. Block 3716 compares the index “i” to the termination value “j”. If the comparison indicates that all relevant snapshots have been searched the current volume value is unchanged from the restoration snapshot and the process returns to 3708. Block 3718 determines if the selected granule has been cached for the selected snapshot. If so the process continues at 3722 replacing the volume granule data with the located cache granule data and continuing to 3708. If the granule is not located in 3718 then block 3720 will increment the snapshot index “i” and continue execution at 3714.
The user experience in restoring the system to a previous snapshot is illustrated by screenshots in
To insure against inadvertent reversions, an initiation sequence preferably is utilized in accordance with preferred embodiments of the present invention wherein a user's intention to perform the reversion operation on the computer system is confirmed prior to such operation. Preferred initiation sequences are disclosed, for example, in copending Witt International patent application serial no. PCT/US02/40106 filed Dec. 16, 2002, and Witt U.S. patent application Ser. Nos. 10/248,425 filed Jan. 18, 2003; 10/248,424 filed Jan. 19, 2003; 10/248,425 filed Jan. 19, 2003; 10/248,426 filed Jan. 19, 2003; 10/248,427 filed Jan. 19, 2003; 10/248,428 filed Jan. 19, 2003; 10/248,429 filed Jan. 19, 2003; and 10/248,430 filed Jan. 19, 2003, each of which is incorporated herein by reference.
Utilization of Snapshots in New and Useful Ways
In view of the systems and methods of managing snapshots as now described in detail herein, and as exemplified by the source code of the U.S. provisional patent application and Appendix A that is incorporated by reference herein, revolutionary benefits and advantages now can be had by utilizing snapshots in many various contexts that, heretofore, simply would not have been practical if not, in fact, impossible. Several such utilizations of snapshots that are enabled by the systems and methods of managing snapshots disclosed herein, including by the incorporated code, are considered to be part of the present invention, and now are described below.
HDD Data History, Virus Protection, and Disaster Recovery
A conventional hard disk drive (HDD) controller, which may be located on a controller board within a computer or within the physical HDD hardware unit itself (hereinafter “HDD Unit”), includes the capability to execute software. Indeed, controller boards and HDD Units now typically when shipped from the manufacturer include their own central processing units (CPU), memory chips, buffers, and the like for executing software for processing reads and write to and from computer readable storage media. Furthermore, the software in these instances is referred to as “firmware” because the software is installed within the memory chips (such as flash RAM memory or ROM) of the controller boards or HDD Units. The firmware executes outside of the environment of the operating system of the computer utilizing the HDD storage and, therefore, is generally protected against alteration by software users of computers accessing the HDD and computer viruses, especially if implemented in ROM. Firmware thus operates “outside of the box” of the operating system. An example of HDD firmware utilized to make complete and incremental backup copies of a logical drive to a secondary logical drive for backup and fail over purposes is disclosed in U.S. patent application Ser. No. 2002/0133747A1, which is incorporated herein by reference.
In accordance with the present invention, computer executable instructions for taking and maintaining snapshots is provided as part of the HDD firmware, such as in a HDD controller board (see
With reference to
Optionally, the HDD Unit 4448 includes a second connector 4416 as shown in
It should be noted that a security device 4406 is provided in association with the HDD controller card 4404 in
In a preferred embodiment, approximately 20% of the HDD capacity is allocated for the finite data storage for preserving snapshot data by the firmware. Accordingly, the data storage for preserving the snapshot data of a 200 gigabyte HDD, which costs only about U.S. $300 today, would include a capacity of approximately 40 gigabytes, leaving 160 gigabytes available to the computer system for storage. Indeed, preferably only 160 gigabytes is presented to the operating system and made accessible. The other 40 gigabytes of data storage allocated for preserving the snapshot data preferably is not presented to the computer operating system.
It is believed that an average use of a computer, such as a desktop for home or business use, results in approximately a quarter megabyte of net changes per day for the entire 160 gigabyte HDD (i.e., there is a quarter megabyte difference on average when the HDD is viewed at day intervals). Preferably, the HDD firmware takes a new snapshot every day at some predetermined time or at some predetermined event. Under this scenario, snapshots can be taken and maintained for each for approximately one hundred and sixty thousand days, or 438 years (assuming the computer continues to be used during this time period). Essentially, a complete history of the state of the computer system as represented by the HDD each day automatically can be retained as a built in function of the HDD! If the snapshots maintained by the firmware are read only, rather than read write, and if the security device in accordance with preferred embodiments as shown, for example, in
First, for instance, as a result of the HDD data history, disaster recovery can be performed by recovering data, files, etc., from any previous day in the life of the HDD Unit. Any daily snapshot throughout the life of the HDD Unit is available as it existed at the snapshot moment on that day. Indeed, the deletion of a file or infection thereof by a computer virus, for example, will not affect that file in any previously taken snapshot; accordingly, that file can be retrieved from a snapshot as it exited on the day prior to its deletion or infection.
Furthermore, the files of the snapshots of the HDD data history themselves can be scanned (remember that each snapshot is represented by a logical container on the base volume presented to the operating system of the computer) to determine when the virus was introduced into the computer system. This is especially helpful when virus definitions are later updated and/or when an antivirus computer program is later installed following infection of the computer system. The antivirus program thus is able to detect a computer virus in the HDD data history so that the computer system can be restored to the immediately previous day. Files and data not infected can also then be retrieved from the snapshots that were taken during the computer infection once the system has been restored to an uninfected state (remember that a reversion to a previous state does not delete, release, or otherwise remove snapshots taken in the intervening days that had followed the day of the state to which the computer is restored).
This extreme HDD data history also provides enormous dividends for forensic investigations, especially by law enforcement or by corporations charged with the responsibility of how their employees conduct themselves electronically. Once a daily snapshot is taken by the HDD firmware, it is as good as “locked” in a data vault and, in preferred embodiments, is unchangeable by any system user or software. The data representing the state of the HDD for each previous day is revealed, including email and accounting information. Furthermore, unless a user is expressly made aware of the snapshot functionality of the HDD firmware, or unless a user is permitted to explore the “snapshot” folder preferably maintained on the root directory of the volume, the snapshots will be taken and maintained seamlessly without the knowledge of the user. Only the computer administrator need know of the snapshots that occur and, preferably with physical possession of the key to the security device, the administrator will know that the snapshots are true and secure.
The same benefits are realized if the HDD Unit is used in a file server, or if the HDD Unit is used as part of network attached storage. For example, forty average users of a 200 gigabyte HDD would each have access to HDD data history representing the state of their data as it existed for each day over a ten year period. In order to protect against physical damage to the HDD Unit, data of the HDD Unit can be periodically backed up in accordance with conventional techniques, including the making of a backup copy of one of the snapshots itself while continued, ongoing access to the HDD is permitted.
In continuing with the HDD data history example, the snapshots can be layered by taking additional snapshots at a different, periodic interval. Accordingly, at the end of each week, a snapshot can be taken of the then current snapshot of that day of the week to comprise the “weekly” snapshot “series” or “collection.” A weekly snapshots series and a monthly snapshot series then can be maintained by the HDD firmware. Presentation of these series to a user would include within a “snapshot” folder on the root directory two subfolders titled, for example, “weekly snapshots” and “daily snapshots.” Within the “weekly snapshots” would appear a list of folders titled with the date of the day comprising the end of the week for each previous week, and within each such folder would appear the directory structure of the base volume in the state as it existed on that day. Within the “daily snapshots” would appear a list of folders titled with the date of each day for the previous days, and within each such folder would appear the directory structure of the base volume in the state as it existed on that day. This layering of the snapshots could further include a series of “monthly snapshots,” a series of “quarterly snapshots,” a series of “yearly snapshots,” and so on and so forth. It should be noted that little additional data storage space would be consumed by taking and maintaining these different series of snapshots.
If desired, the data storage for preserving the snapshots could be managed so as to protect against the unlikely event that the data storage would be consumed to such an extent that the snapshot system would fail. Preferred methods for managing the finite data storage are disclosed, for example, in copending Green U.S. patent application Ser. Nos. 10/248,460; 10/248,461; and 10/248,462, all filed on Jan. 21, 2003, and each of which is incorporated herein by reference.
Accordingly, but for protection against physical damage to the HDD Unit itself, such as damage by fire or a baseball bat, all of the benefits of conventional snapshots and backups are realized without the time and storage capacity constraints by the seamless integration into the HDD firmware of the systems and methods present invention. Indeed, the taking and maintaining of the snapshots is unnoticeable to the casual eye.
Temporal Database Management and Analysis, National Security/Homeland Defense, and Artificial Intelligence
Much academic and industry discussion has been focused in recent years on how to incorporate time as a factor in database management. See, for example, “Implementation Aspects of Temporal Databases,” Kristian Torp, http://www.cs.auc.dk/ndb/phd_projects/torp.html (copyrighted 1998, 2000); “Managing Time in the Data Warehouse,” Dr. Barry Devlin, InfoDB, Volume 11, Number 1 (June 1997); and “It's About Time! Supporting Temporal Data in a Warehouse,” John Bair, InfoDB, Volume 10, Number 1 (February 1996), each of which is incorporated herein by reference.
As recognized by Kristian Torp, for example, multiple versions of data are useful in many application areas such as accounting, budgeting, decision support, financial services, inventory management, medical records, and project scheduling, to name but a few. Temporal relational database management systems (DBMSs) are currently being designed and implemented that add built in support for storing and querying multiple versions of data and represent improvements to conventional relational DBMSs that only provide built in support for one (the current) version of data. Kristian Torp proposes in his thesis techniques for timestamping versions of data in the presence of transactions.
Furthermore, a debate has arisen between whether time should be taken into account by database management programs themselves (the “incorporated” model), or whether time should be taken into account by applications that access the data from database management programs (the “layered” model).
The snapshot method and system of the present invention introduces yet a third, heretofore unknown and otherwise impractical, if not impossible, means for accounting for time as a factor in database management. Indeed, the method of taking and maintaining multiple snapshots inherently takes time into account, as time inherently is a critical factor in managing snapshot data. Thus, by taking and maintaining snapshots of data, each snapshots represents an instance of that data (it's state at that snapshot time) and the series of snapshots represent the evolution of that data. Moreover, the higher the frequency of snapshots, the greater the resolution and less the granularity of the evolution of the data as a function of time. Accordingly by utilizing snapshot technology preferably as provided by the systems and methods of the present invention, non temporal relational database management systems can be snapshot on an ongoing basis, with the combination of all the snapshots data thereof thereby comprising a temporal data store.
Furthermore, within the context of referring to a temporal database, the present invention is considered to provide a temporal data store comprising a plurality of temporal data groups. In this regard, each temporal data group is unique to a point in time and includes one or more snapshots taken at that particular point in time, with the object of each snapshot comprising (1) a logical container, such as a file, a group of files, a volume, or portion of any thereof; or (2) a computer readable storage medium, or any portion thereof. Thus, except in the case where a data group is writable, all data in a dataset necessarily shares in common the characteristic that the data is as it existed at the dataset time point. For example, a snapshot of a first volume at a first time point and a snapshot of a second volume at that same time point, together, may comprise a temporal data group. In juxtaposition, snapshots forming part of a collection or series each is taken at a different time point and, therefore, will not coexist within the same data group as another snapshot of the series, although each snapshot of the series will have in common the same object.
As with multiple versions of data in conventional DBMSs, the temporal data store provided by the present invention efficiently provides multiple versions of data in the form of snapshot series or collections for analysis in many application areas such as accounting, budgeting, decision support, financial services, inventory management, medical records, and project scheduling. Furthermore, neither an incorporated architecture nor a layered architecture is necessary if the snapshot technology is utilized for managing and analyzing the temporal data. A series of snapshots continuously taken of the data suffices, and neither database management programs nor specific applications interfacing with such database management programs need to specifically be rewritten or modified to now account for time as a dimension. Running of the applications in the “current” time while reading the temporal data from the various instances of the data contained within the snapshot folders of the base volume in accordance with the present invention readily provides the solution now sought by so many others for accounting for time as a factor in database management.
Above and beyond providing advantages of conventional DBMSs, the temporal data store of the present invention further provides the ability to conduct multiple “what if” scenarios starting at any snapshot of a data group within a snapshot series. Specifically, because of the additional cache provided in conjunction for each snapshot for writes to the snapshot above and beyond the cache provided for preservation of the snapshot data from the volume, the present invention includes the ability to return to the “pristine” snapshot (original snapshot without write thereto) by simply clearing the write cache. Multiple scenarios thus may be run for each snapshot starting at the same snapshot time (i.e., “temporal juncture” of the various scenarios), and an analysis can be conducted of the results of each scenario and compared to the others in contrasting and drawing intelligence from the different results. In running the different scenarios, different rule sets can be applied to each snapshot for each scenario and within the context of each snapshot folder without altering the current state of the system and without permanently destroying the original snapshot. Moreover, because all snapshots are presented in the current state of the system, “what if” scenarios can be conducted on various, different snapshots in parallel. This ability to utilize snapshot technology to a run “what if” scenario on a snapshot, as well as to return to the pristine snapshot and rerun a different “what if” scenario using a different rule set, all while doing in parallel a similar analysis on other snapshots, provides a heretofore unknown and incredibly powerful analytical tool for data mining and data exploration. Moreover, by considering consecutive snapshots in a series in this analysis, data evolution can also be analyzed from each temporal juncture of the series.
The implications for utilization of the snapshot technology of the present invention in intelligence gathering, especially for counter terrorism and national security interest, are staggering. Currently, the storage capacity required for the ability to run the magnitude of equivalent scenarios provided by the present invention is impractical if not impossible, even for the National Security Administration (or the recently created Department of Homeland Defense). For example, multiple rule sets for data mining and exploration in intelligence gathering can now be applied to snapshots of the data captured by the governmental intelligence agencies and different scenarios for each temporal juncture in snapshot series run in parallel. As a result of the present invention, for each temporal juncture of each snapshot identified for investigation, a system no longer need be restored to its previous state at a temporal juncture, the scenario executed, the system restored again to the same temporal juncture, and the next scenario executed, and so on and so forth. Consequently, the implications are staggering. Snapshots existing on every day between Jan. 1, 2001, and Sep. 11, 2001, of email traffic passing through a particular node of the Internet backbone can be conveniently analyzed under different rules sets and investigative algorithms to determine which would be more effective and what information could have/was known or available within the data archives that might have forewarned authorities to the tragic events that happened on Sep. 11, 2001.
It will also be apparent to those of ordinary skill in the art that the ability to “backtrack” to a previous temporal juncture and execute a different rule set also provides enormous advantages and additional functionality to artificial intelligence.
In summary, revolutionary advancements in data analysis and intelligence can now be had in areas such as medical information analysis (especially patient information analysis); financial analysis, including financials market analysis; communications analysis (such as email correspondence), especially for intelligence pertaining to terrorism and other national security/homeland defense interests; and Internet Archiving and analysis. In each of these examples, the relevant data in the state as it existed for points in time can be readily analyzed online by appropriate algorithms, routines, and programs, especially those utilizing artificial intelligence and backtracking techniques.
While it will now be readily evident that the methods and systems for taking and maintaining snapshots of the present invention far exceeds the mere use of a snapshot for creation of a backup copy onto some backup medium, such use of a snapshot nevertheless remains valid. Thus, in accordance with a feature of the present invention, a snapshot of a volume is represented as a logical drive when a backup of that volume is to be made. Thus, the backup program obtains the data of the snapshot by reading from the logical drive and writing the data read there from onto the backup medium, such as tape. Alternatively, the backup method and system of U.S. patent application Ser. No. 2002/0,133,747A1 is utilized in creating a backup. Moreover, a preferred embodiment of the present invention includes the combination of the backup method and system of U.S. patent application Ser. No. 2002/0,133,747A1 with the inventive snapshot method and system as generally represented by the code of the incorporated provisional patent application and described in detail above. Indeed, the backup may be made by reading not from the base volume itself, but from the most recent snapshot, thereby allowing continuous reads and writes to a base volume during the backup process.
In view of the foregoing detailed description of preferred embodiments of the present invention, it readily will be understood by those persons skilled in the art that the present invention is susceptible of broad utility and application. While various aspects have been described in the context of backup, database, and data analysis uses, the aspects may be useful in other contexts as well. Many embodiments and adaptations of the present invention other than those herein described, as well as many variations, modifications, and equivalent arrangements, will be apparent from or reasonably suggested by the present invention and the foregoing description thereof, without departing from the substance or scope of the present invention. Furthermore, any sequence(s) and/or temporal order of steps of various processes described and claimed herein are those considered to be the best mode contemplated for carrying out the present invention. It should also be understood that, although steps of various processes may be shown and described as being in a preferred sequence or temporal order, the steps of any such processes are not limited to being carried out in any particular sequence or order, absent a specific indication of such to achieve a particular intended result. In most cases, the steps of such processes may be carried out in various different sequences and orders, while still falling within the scope of the present inventions. In addition, some steps may be carried out simultaneously. Accordingly, while the present invention has been described herein in detail in relation to preferred embodiments, it is to be understood that this disclosure is only illustrative and exemplary of the present invention and is made merely for purposes of providing a full and enabling disclosure of the invention. The foregoing disclosure is not intended nor is to be construed to limit the present invention or otherwise to exclude any such other embodiments, adaptations, variations, modifications and equivalent arrangements, the present invention being limited only by the claims appended hereto, or presented in any continuing application, and the equivalents thereof.
Thus, for example, it is contemplated within the scope of the present invention that the finite data storage for preserving snapshot data, while having a fixed allocation in preferred embodiments of the present invention, nevertheless may have a dynamic capacity that “grows” as needed as disclosed, for example, in U.S. Pat. No. 6,473,775, issued Oct. 29, 2002, which is incorporated herein by reference.